Practical Data Lab is a full-service data science company specializing in data analysis and data visualization.
Whether your aim is to reduce expenses, retain customers, increase productivity, or market your success, we can help. Practical Data Lab uses the tools of data science to convert data into insight, insight into action.
With deep experience in diverse statistical methods and subject matter expertise that spans human resources and workforce, education and policy, and many fields in between, we help companies, universities, and nonprofits learn from their data to realize their vision.
Our name is our mission: no-nonsense, custom-tailored, Practical ... for you. Your data. Your product. Your strategy. The insight and action you need to propel your organization forward.
At the most general level, analytics refers to the responsible use of data to derive insight. It can comprise anything from descriptive information, such as counts, means, and medians, which can be generated simply by querying a database, to inferential analyses using, for example, multiple regression or generalized linear models, to predictive analyses using machine learning algorithms such as gradient boosting machines or neural networks, which require specialized statistical software, such as R or Python, to implement. Along with data visualization, analytics comprise the core of the work that we do here at the Practical Data Lab. Our analytics work allows companies, universities, and nonprofits to learn from their data and put that learning into practice.
Data visualization is the visual, as opposed to textual, display of data. Visual display facilitates decision making by allowing key stakeholders to literally see patterns and trends in their data. This approach is especially powerful when integrated with data analysis so that results obtained using more sophisticated analytic methods can be presented in a manner that a diverse array of stakeholders can grasp. At Practical Data Lab, our data visualizations run the gamut from static dashboards, to interactive visualizations where the end user can make her own selections to drill down to greater and greater levels of detail, to professionally executed video storytelling, that communicates success to external stakeholders with engaging 2D/3D animation & motion graphics.
It's an organization’s strategy for using its existing data assets and acquiring new ones to advance its goals. A comprehensive data strategy encompasses not only data use and acquisition, but also storage, management, integration, transfer, governance, and security, as well as all of the human capital, software, and hardware and networking infrastructure to support those activities. Practical Data Lab helps companies construct comprehensive data strategies by tailoring Data Strategy Audits and Data Strategy Plans to each client. Keep an eye out for our forthcoming White Paper on Data Strategy for Physicians’ Offices: Using Data to Improve Health Care Outcomes.
Does your organization store its data in a bunch of spreadsheets that can’t talk to each other or in separate databases that can’t be joined? This is a problem of data integration, actually, lack of data integration. Data Integration is the process of combining disconnected data assets so that an organization can use them to advance its goals. Data integration is a perfect case of the whole being greater than the sum of the parts: while it may be possible to learn only a little from a single, isolated set of data, much may be learned when it is combined with other data assets. Though not always possible, a combination of deterministic and fuzzy matching methods can often successfully join previously disconnected data. Practical Data Lab can assess your data’s suitability for integration and then integrate it when feasible. Take a look at our case study: Matching Data without a Common Identifier.
Ask ten data scientists what big data is and you’re sure to get ten different answers. Beyond the obvious point that big data means lots of data, big data stands apart from traditional small and medium data for the sheer amount of it— it can’t be stored on a single server; the speed with which organizations acquire it; and the multiplicity of types of data — e.g., sensor, video, audio, unstructured text, etc. — each of which presents unique management and analysis challenges. Practical Data Lab helps clients leverage their big data collections to support customer retention, expense reduction, and other key objectives.
Evaluation is the assessment of the effectiveness of any corporate or nonprofit initiative or strategy. It involves not only using the proper statistical algorithms but as (or actually more) importantly designing the analysis in such a way as to generate the most accurate answer possible. Using the logic of evaluation and causal inference, Practical Data Lab helps clients assess how effective their strategies are at helping to retain customers, reduce expenses, and increase productivity.
How we work reflects our passion for data and our personalities.
Producing top quality data products requires a thorough understanding of each client’s goals, intimate knowledge of their data, and use of the most appropriate data science tools for the job. For this reason, every project we undertake at the Practical Data Lab begins with Data Discovery. During Data Discovery, we learn about both our client’s needs and goals and as well as all the ins and outs of your data. We ask, for example, What is the meaning of each field of data? How were the data generated or collected? How are the data stored and maintained? We leverage our client’s deep knowledge of their own data to develop a realistic plan that lays out what we can learn from the data, the methods we will use to generate insight, and the products we will deliver. We have found that this collaborative approach allows us to produce the results that best meet our clients’ needs.
With you: We emphasize personal relationships, so that we treat our clients as people, not account numbers. With your data: We immerse ourselves in every detail so that we deliver exactly what you want.
Detail Obsessive Data Dorks.
And proud of it! You’d have hated having any of us as your sixth grade grammar teacher, but you’ll love us as we dot and cross every i and t, respectively :), for you. Your data will thank you.
We listen. We learn. Our sole focus is getting the job done to meet your needs, not fuel our egos. You will thank yourself.
Communication and transparency comprise our secret sauce. We need to understand your needs, your vision, your specific objectives.
Of complexity into clarity. Of ideas into analysis. Of the messiest %^$&#@! spreadsheet ever seen into data, insight, and action.
We take our work extremely seriously but try to maintain a light-hearted atmosphere so as to make the process as painless as possible for our clients.
William Mabe, PhD: Chief Data Scientist
Practical Data Lab is led by our Chief Data Scientist, Bill Mabe. Bill has been nerding out on data since 1990, the year he took his first data science courses — back then, they were just called statistics — at the University of Maryland and had to program all his jobs in SAS and submit them to the mainframe.
He earned his PhD in Political Science at Rutgers University in 2007, taking data science courses in the Rutgers political science and economics departments and at ICPSR at the University of Michigan. Around that time, he became hooked on R and never looked back at SAS or Stata.
In 2003, he began a 14-year stint as a researcher at the John J. Heldrich Center for Workforce Development in the Bloustein School for Planning and Public Policy at Rutgers. There, he managed and analyzed large data sets, designed and implemented predictive modeling projects, built longitudinal databases, produced data visualizations, and conducted qualitative and quantitative evaluations of over 50 different education and workforce programs.
For the last six plus years at the Center, Bill served as the Director of Research and Evaluation in which capacity he set the strategic direction for the Center's research division, managed an annual budget of $1.3 million, and supervised a staff of 10 full- and part-time researchers.
In 2011, Bill earned the distinction of Faculty Fellow at the Bloustein School and in 2016 taught the Bloustein School’s Big Data Analytics graduate seminar.